Skip to main content
NCBI home page
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2016 Dec 12;113(52):14887–14892. doi: 10.1073/pnas.1612628113

Accomplishing simple, solubility-based separations of rare earth elements with complexes bearing size-sensitive molecular apertures

Justin A Bogart a, Bren E Cole a, Michael A Boreen a, Connor A Lippincott a, Brian C Manor a, Patrick J Carroll a, Eric J Schelter a,1
PMCID: PMC5206573  PMID: 27956636

Significance

Rare earth metals, La–Lu, Sc, and Y, are essential components of electronic materials and permanent magnets in diverse technologies. But, their mining and separations chemistry are unsustainable and plagued with supply problems. Recycling of consumer materials containing rare earths is a promising new source of these critical materials but similarly requires efficient separations. We report the use of a tripodal hydroxylaminato ligand, TriNOx3–, with rare earth cations that enable fast, efficient separations through a simple leaching process. This approach is expected to contribute to a new method for recycling postconsumer materials including lighting phosphors and magnets, to diversify and improve the sustainability of the rare earth metal supply chain.

Keywords: rare earth elements, lanthanides, separations, critical materials, coordination chemistry

Abstract

Rare earth (RE) metals are critical components of electronic materials and permanent magnets. Recycling of consumer materials is a promising new source of rare REs. To incentivize recycling, there is a clear need for the development of simple methods for targeted separations of mixtures of RE metal salts. Metal complexes of a tripodal hydroxylaminato ligand, TriNOx3–, featured a size-sensitive aperture formed of its three η2-(N,O) ligand arms. Exposure of cations in the aperture induced a self-associative equilibrium comprising RE(TriNOx)THF and [RE(TriNOx)]2 species. Differences in the equilibrium constants Kdimer for early and late metals enabled simple separations through leaching. Separations were performed on RE1/RE2 mixtures, where RE1 = La–Sm and RE2 = Gd–Lu, with emphasis on Eu/Y separations for potential applications in the recycling of phosphor waste from compact fluorescent light bulbs. Using the leaching method, separations factors approaching 2,000 were obtained for early–late RE combinations. Following solvent optimization, >95% pure samples of Eu were obtained with a 67% recovery for the technologically relevant Eu/Y separation.

Rare earth elements, La–Lu, Y, and Sc, have applications in many technologies, including permanent magnets (1–5), NiMH batteries (6–10), and lamp phosphors (11–14). Limitations associated with their beneficiation and separations, especially their solvent-, waste-, and energy intensities, have contributed to the concentration of suppliers in the People’s Republic of China. Supply risks for these elements have emerged, particularly in the face of current and growing demand in the next 20 y (15, 16). Because the global marketplace for these elements is dominated by a single source (17), prices for primary rare earth (RE) materials are volatile (18). As a result, the US Department of Energy has classified many of these elements as “critical” (19). There is a clear need to find potential new supplies of these elements.

Recent life cycle assessments have indicated that recycling of consumer materials is a promising alternative to conventional production processes (20). Despite this assertion, as recently as 2011, less than 1% of RE-containing materials were being recycled (21). These low recycling rates stem from a combination of sporadic collection procedures and lack of efficient separations and preprocessing steps (22–27).

To contribute to incentivizing the “urban mining” of RE-containing materials, we recently initiated efforts toward new, simplified methods in RE separations (28). Our initial work focused on the separation of neodymium (Nd) and dysprosium (Dy), two key components of neomagnets (Nd2Fe14B). We disclosed the development of the tripodal nitroxide ligand, [((2-tBuNO)C6H4CH2)3N]3− (TriNOx3–), which induced a self-association equilibrium between monomeric Nd(TriNOx)THF/dimeric [Nd(TriNOx)]2 species. The position of this equilibrium was found to be strongly dependent on the size of the RE cation. We showed proof of concept that differences in the self-association equilibrium constants between Nd and Dy could be exploited to achieve 95% pure materials through a simple leaching method with benzene or toluene. It was of interest to extend this method to other RE mixtures, in particular those present in phosphor waste streams.

RE elements used in phosphors for compact fluorescent light bulbs represent 32% of RE market share, comparable to the 38% for permanent magnetic materials (21). Lamp phosphor waste consists of the non–RE-containing halophosphate, (Sr,Ca)10(PO4)6(Cl,F)2:Sb3+,Mn2+ (HALO, 40–50%) and RE-containing components, Y2O3:Eu3+ (YOX, 20%), LaPO4:Ce3+,Tb3+ (LAP, 6–7%), BaMgAl10O17:Eu2+ (BAM, 5%), and small quantities of other phosphors such as (Ce,Tb)MgAl11O19 (CAT) and (Gd,Mg)B5O10:Ce3+,Tb3+ (29). Of these, YOX has the highest intrinsic value because it contains 80 wt % of the total RE content of the phosphor waste and is composed exclusively of the two critical RE elements, Y and Eu. Recent developments in the recycling of phosphor materials have largely focused on the recovery of Eu and Y from YOX (11, 30–38). Dupont and Binnemans recently reported on the selective dissolution of YOX and recovery of yttrium and europium using the functionalized ionic liquid betainium bis(trifluoromethylsulfonyl)imide, [Hbet][Tf2N] (29). Currently, however, only one industrially applied recycling process exists for lamp phosphor waste (29, 39). Solvent extraction of the resulting Eu- and Y-containing concentrates provides high-purity, recycled materials. The main drawbacks of the commercial process are that it involves costly acidic and basic washes at relatively high temperatures to fully disintegrate the phosphors, followed by a manifold of energy-, solvent-, and waste-intensive liquid–liquid extraction steps to purify the individual REs (29). Therefore, recycling of REs from phosphor waste could equally benefit from, simpler, more efficient separations techniques.

Implementation of a similar separations method for Eu/Y mixtures based on the TriNOx3– system necessitates a more complete understanding of the coordination chemistry of the TriNOx3– ligand developed by our group across the series of RE cations. In particular, it was of interest to determine the response of the key self-association equilibrium constants with respect to changes in metal ionic radius. We hypothesized that understanding that relationship would inform on the overall performance of the system, namely through separations factors for pairs of metal cations, SRE1/RE2.

Herein we report the syntheses of RE(TriNOx)THF (1-RE), where RE = Pr, Sm–Tb, Ho–Lu, which completes the series of complexes excepting Pm and Sc. The dimeric [RE(TriNOx)]2 complexes (2-RE), where RE = Pr and Sm, are also reported. Determination of the self-association equilibrium constants, Kdimer, for RE = La–Sm using 1H NMR titration experiments is disclosed along with those values of Kdimer for RE = Eu–Lu, and Y from extrapolation of results for the early metals. The measured and estimated Kdimer show an estimated 1011-fold decrease across the series. A complete matrix of 54 separations factors, SRE1/RE2, is provided for the early/late RE combinations, determined by 1H NMR spectroscopy experiments and inductively coupled plasma optical emission spectroscopy (ICP-OES) data on the separated fractions. We expect that the results disclosed here will aid in the development of improved methods for separating Eu/Y mixtures with potential applications in the recycling of lamp phosphor waste, as well as other RE materials recycling schemes.

Results and Discussion

Synthesis and Characterization.

The series of 1-RE complexes, RE = Pr, Sm–Tb, Ho–Lu, was synthesized using procedures previously published by us for RE = La, Ce, Nd, Dy, and Y (28, 40). Method A (Scheme 1) involved layering hexanes solutions of the tris[N,N-bis(trimethylsilyl)amide] reagents, RE[N(SiMe3)2]3, onto THF solutions of protonated H3TriNOx and isolating the X-ray quality crystals. Method B involved reacting the RE triflate salts, RE(OTf)3, with 1 equiv of H3TriNOx, and 3 equiv of K[N(SiMe3)2] in THF. X-ray quality crystals of the crude 1-RE formed by this route were grown by layering THF onto saturated dichloromethane solutions. The low solubility of 1-RE, RE = Tm–Lu, in dichloromethane and THF/hexanes solutions, however, prevented the isolation of crystals using these methods. Thus, X-ray quality crystals of these complexes were formed by layering a precooled THF solution of K[N(SiMe3)2] onto a precooled THF solution of RE(OTf)3 and H3TriNOx and letting the reaction set undisturbed at –25 °C for 14 h. The X-ray structure of representative 1-La is shown in Fig. 1 and salient structural metrics for 1-RE are provided in Table 1.

Scheme 1.

Scheme 1.

Open in a new tab

Syntheses of 1-RE from RE[N(SiMe)3]3 and H3TriNOx.

Fig. 1.

Fig. 1.

Open in a new tab

Thermal ellipsoid plots of (Top) 1-La and (Bottom) 2-La shown at 30% probability. Interstitial solvent molecules and hydrogen atoms have been omitted for clarity. tert-butyl groups are depicted using a wireframe model.

Table 1.

Structural metrics, %Vbur, and Kdimer for the series of 1-RE complexes

Complex N–O RE–O RE–N RE–OTHF %Vbur Kdimer Ref.
1-La 1.441(2) 2.3219(16) 2.6022(17) 2.595(6) 78.6 3.2 ± 0.9 × 103 28
1-Ce 1.433(2) 2.2921(18) 2.581(2) 2.577(6) 79.1 2.4 ± 0.2 × 102 40
1-Pr 1.424(3) 2.270(2) 2.569(3) 2.573(8) 79.6 1.2 ± 0.1 × 101 This work
1-Nd 1.420(4) 2.260(3) 2.554(4) 2.546(9) 79.9 2.4 ± 0.2 28
1-Sm 1.429(3) 2.236(3) 2.547(3) 2.545(10) 80.4 2.4 ± 0.9 × 10−2 This work
1-Eu 1.438(2) 2.2270(16) 2.5266(19) 2.506(5) 80.9 (3.6 × 10−3) This work
1-Gd 1.437(4) 2.220(3) 2.529(4) 2.530(10) 81.0 (5.6 × 10−4) This work
1-Tb 1.444(2) 2.2054(18) 2.514(2) 2.501(6) 81.3 (8.7 × 10−5) This work
1-Dy 1.424(4) 2.180(3) 2.508(3) 2.487(10) 81.3 (1.3 × 10−5) 28
1-Y 1.4429(17) 2.1784(13) 2.4966(15) 2.469(3) 81.7 (4.2 × 10−6) 28
1-Ho 1.433(5) 2.189(4) 2.506(5) 2.509(10) 81.5 (2.4 × 10−6) This work
1-Er 1.439(4) 2.162(3) 2.487(4) 2.484(10) 82.3 (4.9 × 10−7) This work
1-Tm 1.446(5) 2.151(4) 2.483(6) 2.477(11) 82.5 (1.2 × 10−7) This work
1-Yb 1.441(5) 2.151(4) 2.474(5) 2.447(7) 82.3 (3.2 × 10−8) This work
1-Lu 1.452(4) 2.134(4) 2.480(5) 2.428(8) 82.6 (1.0 × 10−8) This work
Open in a new tab

The metrics obtained from the X-ray data for the 1-RE complexes indicated that they were isostructural across the series. The η2-(N,O) bonding mode of the three hydroxylaminato arms was retained, and each complex contained a single molecule of THF coordinated to the central metal cation. The η2-(N,O) bonding mode had been observed previously by us and others in the RE(TriNOx)THF (RE = La, Ce, Nd, Dy, and Y) (28, 40), [RE[1,3-(2-(tBuNO)C6H4)2-C6H4]2]0/1– (RE = La, Ce, Pr, Y, and Tb) (41, 42), and Ce[η2-ON(tBu)(2-OMe-5-tBu-C6H3)]4 complexes (43), but was contrasted to the κ2-(Npyr,O) bonding mode observed in the tetrakis pyridyl hydroxylaminato lanthanide complexes, [Ln(2-(tBuNO)py)4]0/1–, Ln = La or Ce (44, 45). As shown in Table 1, there was a steady decrease in the metal–ligand bond lengths across the series, consistent with the decrease in ionic radii of the central metal cations. This decrease resulted in a gradual closing of the molecular aperture formed by the three oxygen atoms of the TriNOx3– ligand, as similarly indicated by the increase in percent buried volume (%Vbur) from 78.6 for La to 82.6 for Lu (Fig. 2).

Fig. 2.

Fig. 2.

Open in a new tab

Spacefill diagrams of 4.1-La (Left) and 4.1-Lu (Right) with the THF molecules removed, highlighting the decrease in exposure of the central metal cation.

In order to gain insight into the speciations of these complexes, 1H NMR spectroscopy experiments were performed. Again, the C3 symmetry of these complexes was maintained in d5-pyr. In C6D6 solutions, however, additional resonances for C2 symmetric dimeric structures were observed in solutions of 1-RE, RE = La–Eu. These dimeric [RE(TriNOx)]2 complexes (2-RE) could be isolated for RE = La–Sm by dissolving the respective 1-RE species in toluene and removing the solvent under reduced pressure. X-ray quality crystals of isostructural 2-RE, RE = La–Sm, were grown by layering hexanes onto saturated dichloromethane or toluene solutions of their respective 1-RE complexes or cooling saturated Et2O solutions of 1-RE to –25 °C. The presence of only minimal amounts of dimer in solutions of the 1-Eu species prevented the isolation and crystallization of the putative 2-Eu species.

The structural metrics obtained for the 2-RE complexes show similar features to the isolated monomeric species. As an example, the average N–O bond length of 1.448(4) Å in 2-La (Fig. 1, Bottom) compares well with 1.441(2) Å in 1-La (Fig. 1, Top). The η2-(N,O) La–O bonds in 2-La [2.304(3) Å are slightly shorter than the average hydroxylaminato La–O bond length [2.3219(16) Å] in 1-La, whereas the La–N bonds in 2-La [2.619(3) Å] are longer than those in 1-La [2.6022(17) Å]. The 2-RE structure features one hydroxylaminato arm from each La(TriNOx) fragment binding both La centers in a μ2-O fashion. The bridging hydroxylaminato arm features a longer La–N bond [2.747(3) Å] and two longer La–O bonds [2.418(2), 2.508(2) Å]. Notably, there was a steady decrease in the RE(1)–O(1) bond distances [2.418(2)–2.305(7) Å] across the series RE = La–Sm, consistent with the decrease in ionic radius, whereas the bridging RE(2)–O(1) bond distances did not trend.

Determination of Kdimer.

The presence of appreciable amounts of 2-RE in benzene solutions of 1-RE, RE = La–Sm, allowed for the estimation of the self-association equilibrium constants for these complexes using 1H NMR spectroscopy. C6D6 solutions of 2-RE were titrated with THF, and the equilibrium concentrations of 1-RE and 2-RE were calculated against an internal ferrocene reference. This method had been used previously by us to determine the magnitude of the self-association equilibrium constant, 2.4 ± 0.2, for 1-Nd (28). The results of these experiments indicated that the equilibrium constants decreased by an order of magnitude between adjacent RE ions as the series was traversed from larger to smaller ions. We attributed this decrease in self-association equilibrium constants across the series to the closure of the molecular aperture formed by the oxygen atoms of the TriNOx3– ligand, which increased the steric clash to approach of a second molecule of RE(TriNOx). This hypothesis was supported by the observation that linear correlations to ionic radius of the central metal cation could be made for both %Vbur and the logarithm of the self-association equilibrium constants, log Kdimer (Figs. 3 and 4).

Fig. 3.

Fig. 3.

Open in a new tab

Correlation between relative buried volume versus ionic radius of the central metal cation for the series of 1-RE complexes.

Fig. 4.

Fig. 4.

Open in a new tab

Correlation between the logarithm of the experimentally determined self-association equilibrium constants, log Kdimer, for the series of 1-RE complexes, RE = La–Sm (red circles) and extrapolated log Kdimer values for the 1-RE complexes, RE = Eu–Lu (blue squares).

The linear correlation of log Kdimer to metal cation radius allowed for the estimation of the dimerization equilibrium constants for the later cations 1-RE, RE = Eu–Lu. According to these predictions, there was an ∼1011-fold decrease in the self-association equilibrium constants across the series as a result of the 0.18-Å decrease in ionic radii. As with our findings for Nd/Dy, the high sensitivity of the dimerization equilibrium constants to small changes in ionic radii across the series indicated targeted separations of the RE ions would be feasible. Whereas no mixed dimer formation was observed in the Nd/Dy mixtures as determined by the lack of exchange observed in the 2D 1H EXSY NMR spectroscopy experiments (28), we expected mixed dimer formation in the early/early RE combinations. Indeed, in C6D6 solutions of 1-La and 1-Ce mixtures, resonances corresponding to 2-La, 2-Ce, and a species postulated to be the mixed La/Ce dimer, were observed (SI Appendix). As such, early/late RE combinations where the formation of mixed dimers would be unfavorable were explored for targeted separations.

Targeted RE Separations.

The success of the separations method for purifying Nd/Dy mixtures resulted from the large difference in propensity to form dimeric species in benzene solutions between 1-Nd (Kdimer = 2.4 ± 0.2) and 1-Dy (Kdimer ∼1.3 × 10−5). A similar difference in propensity to form dimeric species in benzene solutions was estimated between 1-Eu (Kdimer ∼3.6 × 10−3) and 1-Y (Kdimer ∼4.2 × 10−6). The 103-fold decrease in magnitude of Kdimer between 1-Eu and 1-Y was promising for targeted separations of these ions. These separations necessitated a more complete understanding of the separations method based on the TriNOx3–-type ligand system across the complete series of early/late RE combinations. As such, separations were performed on mixtures of RE1/RE2(TriNOx)THF combinations, RE1 = La–Eu, RE2 = Gd–Lu, and Y. These mixtures were formed by allowing 50:50 mixtures of RE1/RE2(OTf)3 salts to react with 2 equiv H3TriNOx and 6 equiv. of K[N(SiMe3)2] in THF.

Leaching of the solid RE1/RE2(TriNOx)THF with 4 mL of benzene followed by a wash of 1 mL of benzene produced filtrates enriched with the larger RE1 and solids enriched with the smaller RE2. The molar ratios of the two RE metals in the filtrate and solid fractions were determined by 1H NMR spectroscopy. ICP-OES spectroscopies were also performed for select cases to validate the NMR results and in cases where the paramagnetism of the RE cations prevented outright the measurement of the separations factors due to relaxation and broadening.

From these data, the separations factors, SRE1/RE2, could be computed according to the equation SRE1/RE2 = Dsolid ⋅ Dfiltrate. The enrichment factors, D, were determined as the molar ratio of the enriched RE species to the minor RE species in the sample. As such, Dfiltrate was calculated as nRE1/nRE2 and Dsolid as nRE2/nRE1, where n is the molar ratio of RE species by 1H NMR or ICP-OES. The results of the RE1/RE2 separations are tabulated in Table 2 and depicted in Fig. 5. The enrichment factors for the solid and filtrate factors were also determined and tabulated as Supporting Information (SI Appendix). (For an example calculation of enrichment and separations factors, see SI Appendix).

Table 2.

Separations factors, SRE1/RE2, for the series of RE1/RE2

RE1 RE2
Gd Tb Dy Y Ho Er Tm Yb Lu
La (7.32)* 91.0 182(54.4) 341 834 176 1,194 239 90.9
Ce (58.4) 60.2 495(276) 122 337 260 1,942 138 520
Pr (20.2) 457 355(224) 485 920 226 259 286 610
Nd (64.0) 133(179) 302(359)† 319(450) 532(1,089)‡ 1,222(1,220) 29.4(38.5) 181(175) 39.6(660)‡
Sm (2.72) 35.8 55.6 61.6 119 46.4 36.6 57.3 106
Eu (3.17) 13.6 38.2 28.4 (39.2) 77.4 22.0 72.4 26.9 10.7
Open in a new tab
*

Values in parentheses obtained from the ICP-OES data.

†

See ref. 28.

‡

Differences between the ICP-OES and NMR results are discussed in the text.

Fig. 5.

Fig. 5.

Open in a new tab

Bar graph of the separations factors, SRE1/RE2, for the early/late RE combinations.

In general, the separations factors, SRE1/RE2, increased from larger to smaller RE2 ions across each row for a given RE1 ion. Furthermore, down each column for a given RE2 ion, the separations factors decreased from larger to smaller RE1 ions. These trends were expected due to the greater tendency of the larger RE ions to form higher concentrations of dimeric species and be extracted into the benzene solutions. Surprisingly, the trends were not systematic and breaks in the separation factors were observed, in particular for the very late RE2 ions in each row and the very early RE1 ions in each column. These NMR data were corroborated by ICP-OES data on the Nd/late RE and early RE/Dy separations, which showed good agreement with the 1H NMR results.

The discrepancy in the values for SNd/Lu could be explained by the insolubility of 1-Lu in d5-pyr, which caused an artificially low result for the enrichment factor of the solid fraction. Therefore, the 1H NMR values for the RE1/Lu combinations provided a lower limit for the SRE1/Lu separations factors. Similarly, the discrepancy in the values for SNd/Ho could be explained by error in the integration of 1-Ho in the solid fraction due to the severe paramagnetic relaxation of the signal due to the Ho metal center. The enrichment factor of the filtrate fraction in this case, however, was accurately determined by 1H NMR spectroscopy and matched the ICP-OES value.

The results from the separations based on the TriNOx3–-system were compared with the RE-HCl-HDEHP- and RE-HNO3-Cyanex 302-based separations used industrially (46, 47). Unlike the TriNOx3–-system, the separations factors for both the HDEHP- and the Cyanex 302 systems increased linearly as the difference in ionic radii between the two RE ions increased with no breaks to the trend. On average, the separations factors for the TriNOx3– system were approximately five times larger than those for the HDEHP system. In contrast, the separations factors for the TriNOx3– system were comparable to those of the Cyanex 302 system for RE1 = Pr, Nd, Sm, Eu, and RE2 = Tb, Dy, Ho. The separations factors for the TriNOx3– system, however, deviated from those of the Cyanex 302-system for RE1 = La, Ce and/or RE2 = Y, Er–Lu. As such, the data showed separations factors for the TriNOx3– system that were ∼0.4 times smaller than those for the Cyanex 302-system on average.

We hypothesized that the decreases in separations factors of the very early RE1 ions for a given RE2 ion in the case of the TriNOx3− system were results of increased mixed dimer formation. For example, for RE2 = Dy, the separations factors decreased from the maximum value of SNd/Dy = 359 for RE1 = Nd to SLa/Dy = 54.4 for RE1 = La. These decreases in separations factors from RE1 = Nd to La followed the trend expected from increased mixed dimer formation occurring in solutions of the larger RE1 cations, where the greater exposure of the larger cation would allow for the approach of a greater amount of RE2(TriNOx) species. This would result in decreased enrichment factors for both the solid and solution phases.

Similarly, we hypothesized that the decreases in separations factors of the very late RE2 ions for a given RE1 ion were the result of increased free THF concentrations in solutions of the latter due to dissociation of the weakly bound THF molecules. For example, for RE1 = Nd, the separations factors reached a maximum of SNd/Er = 1,220 for RE2 = Er before decreasing to SNd/Tm = 38.5 for RE2 = Tm. Despite the separations factors increasing again to SNd/Lu = 660 for RE2 = Lu due to the lower propensity of dimer formation for Lu, the values did not exceed the value achieved for SNd/Er. Again, these decreases in separations factors follow the trend expected for increased amounts of free THF in solutions of the later RE2 ions as the series of RE2 were traversed from Er to Lu. Dissociation of THF from RE2(TriNOx)THF, and increased concentrations of free THF in solution, would shift the equilibrium between RE1(TriNOx)THF/[RE1(TriNOx)]2 toward monomer. This shift in equilibrium position would result in decreased amounts of RE1 leached into the filtrate fraction, which would in turn decrease the enrichment factors of both the solid and filtrate fractions. To test this hypothesis, a relative estimate of the RE–OTHF bond enthalpies was measured by thermogravimetric analysis (TGA) analysis on 1-RE, RE = La, Nd, Eu, Dy, Y, and Tm (Table 3).

Table 3.

Weight decrease and THF dissociation temperature for 1-RE (RE = La, Nd, Eu, Dy, Y, Tm)

Complex Weight decrease, % THF dissociation, °C
1-La 7.7, 10.8 89, 168
1-Nd 6.3, 12.2 88, 153
1-Eu 16.4 118
1-Dy 15.8 85
1-Y 9.3 85
1-Tm 6.5 75
Open in a new tab

In the case of 1-La, two small decreases in weight percent of 7.7% and 10.8% were observed at 89 °C and 168 °C, respectively, before the onset of decomposition was observed as indicated by the large decrease in weight percent at 220 °C (Table 3). The small decreases in weight percent were assigned as the loss of interstitial and bound THF, respectively. The TGA of 1-Nd was similar to that of 1-La except the decrease of 12.2% for the bound THF at 153 °C. In contrast, only one small decrease in weight percent of 16.4% at 118 °C was observed before complex decomposition in the case of 1-Eu. The TGA data for 1-Dy, 1-Y, and 1-Tm were similar to that of 1-Eu with decreases of 15.8%, 9.3%, and 6.5% at 85 °C, 85 °C, and 75 °C, respectively. These peaks corresponded to the dissociation of both interstitial and bound THF, which occurred at unresolved temperatures in these cases. Taken together, these results indicated that there was a systematic decrease in Ln–OTHF bond strength across the series, which supported the hypothesis that secondary equilibria involving the dissociation of bound THF complicated the separations processes.

Solvent Optimization.

The separations factor of 28.4 for the Eu/Y mixtures and operational simplicity of the method had potential implications in the recycling of RE-containing phosphor materials. Furthermore, the high enrichment factor of 18.3 for europium in the filtrate fraction suggested that the small separations factor was a result of limited solubility of 1-Eu in benzene solution but that pure samples could be obtained with this separations method. We hypothesized that the use of polar, noncoordinating solvents would lead to an increase in the solubility of 1-Eu. As such, we attempted the separation of Eu/Y mixtures in a number of noncoordinating solvents with variable dielectric constants (Table 4) (48). Whereas no correlation was found between dielectric constant and separation factor, the low solid enrichment factor and high filtrate enrichment factor for diethyl ether were of interest. Increasing the volume of ether from 4 to 15 mL resulted in a separation factor of 62.2, whereas stirring the solids in 15 mL ether twice and combining the filtrate portions resulted in a separation factor of 189. This value is higher than the corresponding values for Cyanex-302 (SEu/Y = 116) and HDEHP (SEu/Dy = 3.83, SEu/Ho = 4.74) (46, 47). Thus, a simple optimization of solvent washes improves targeted separations of Eu/Y.

Table 4.

Enrichment factors, Dfiltrate and Dsolid, and separations factors, SEu/Y, for Eu/Y mixtures in solvents with dielectric constant, εr

εr Dfiltrate Dsolid SEu/Y
Benzene 2.2825 12.1 1.69 20.4
Toluene 2.379 9.84 1.68 16.6
Fluorobenzene 5.465 4.78 9.48 45.3
Chlorobenzene 5.6895 14.1 2.65 37.6
Dichloromethane 8.93 0.927 1.16 1.08
α,α,α-Trifluorotoluene 9.22 3.65 0.94 3.42
Diethyl ether 4.2666 3.47 1.03 3.57
Diethyl ether (15 mL) (×2) 4.2666 26.7 7.07 189
Open in a new tab

Conclusions

Excepting Pm and Sc, a complete series of monomeric RE(TriNOx)THF complexes was synthesized. X-ray crystallography confirmed that these were isostructural in the solid state with each maintaining the binding of a single THF molecule in the apical position. 1H NMR spectroscopy in C6D6 solutions confirmed the formation of dimeric [RE(TriNOx)]2 species for RE = La–Eu, whereas no dimeric species were observed for RE = Tb–Lu, and Y. These dimeric species were isolable for RE = La–Sm and their structures were confirmed by X-ray crystallography. These results suggested that a monomer–dimer “break” occurred at Gd in benzene solutions.

Self-association equilibrium constants for these species were determined by 1H NMR spectroscopy titration experiments on isolable [RE(TriNOx)]2 complexes, RE = La–Sm, and from these data the equilibrium constants for RE = Gd–Lu, and Y could be extrapolated. An estimated 1011-fold decrease in Kdimer was extrapolated, which showed the high sensitivity of the self-association process to slight changes in RE cation radius.

Extension of the previously disclosed separations method for Nd/Dy mixtures based on the TriNOx3– system was performed on the early/late RE combinations, with particular focus on Eu and Y, critical components of phosphor materials in fluorescent light bulbs. In general, the separations factors increased as the difference in ionic radii of the pair of RE ions increased. In the extreme cases with the very large early RE ions and the very small late RE ions, decreases in the separations factors were observed. We attributed these observations to increased amounts of mixed dimer formation with the early RE ions and increased amounts of dissociated THF in solutions of the late RE ions, establishing the limitations of this separations system.

Whereas a moderate separations factor, SEu/Y = 28.4, was observed in the Eu/Y case, the large enrichment factor for the filtrate fraction of 18.3 and moderate recovery of 25% for Eu suggested that this method would be viable for Eu/Y separations. A more than sixfold improvement of the separations factor to 189 was obtained through adjusting the solvent to diethyl ether, indicating simple solvent optimization can have a major impact on improving system performance. Further improvements to these separations are expected through ligand modification, choice of solvent, and optimization of conditions for each RE1/RE2 combination. These studies are ongoing.

Materials and Methods

Methods.

Unless otherwise noted, all reactions and manipulations were performed under an inert atmosphere (N2) using standard Schlenk techniques or in a drybox equipped with a molecular sieves 13X/Q5 Cu–0226S catalyst purifier system. Glassware was oven-dried for 3 h at 150 °C before use.

Materials.

THF, dimethoxyethane, diethyl ether, dichloromethane, toluene, hexanes, and pentane were purchased from Fisher Scientific. All solvents were sparged for 20 min with dry N2 and dried using a commercial two-column solvent purification system.

Supplementary Material

Supplementary File

Acknowledgments

We thank Dr. David Weinberger for preliminary TGA work. E.J.S. acknowledges the US Department of Energy, Office of Science, Early Career Research Program (Grant DE-SC0006518), the Research Corporation for Science Advancement (Cottrell Scholar Award to E.J.S.), and the University of Pennsylvania for financial support of this work. B.E.C. thanks the NSF Graduate Research Fellowship program for support. This work used the Extreme Science and Engineering Discovery Environment, which is supported by the National Science Foundation Grant ACI-1053575.

Footnotes

Conflict of interest statement: The authors declare the following competing financial interest: Intellectual property pertaining to the technology described in this article is covered by International Patent Application no. PCT/US2015/042703.

This article is a PNAS Direct Submission.

Data deposition: The crystallography atomic coordinates and structure factors have been deposited in the Cambridge Structural Database (CSD), https://www.ccdc.cam.ac.uk/solutions/csd-system/components/csd/ (accession nos. 1513324–1513336).

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1612628113/-/DCSupplemental.

References

  • 1.Sagawa M, Fujimura S, Yamamoto H, Matsuura Y, Hiraga K. Permanent magnet materials based on the rare earth-iron-boron tetragonal compounds. IEEE Trans Magn. 1984;20(5):1584–1589. [Google Scholar]
  • 2.Goll D, Kronmüller H. High-performance permanent magnets. Naturwissenschaften. 2000;87(10):423–438. doi: 10.1007/s001140050755. [DOI] [PubMed] [Google Scholar]
  • 3.Smith Stegen K. Heavy rare earths, permanent magnets, and renewable energies: An imminent crisis. Energy Policy. 2015;79:1–8. [Google Scholar]
  • 4.Dent PC. Rare earth elements and permanent magnets (invited) J Appl Phys. 2012;111(7):07A721. [Google Scholar]
  • 5.Gutfleisch O, et al. Magnetic materials and devices for the 21st century: Stronger, lighter, and more energy efficient. Adv Mater. 2011;23(7):821–842. doi: 10.1002/adma.201002180. [DOI] [PubMed] [Google Scholar]
  • 6.Scheffler GL, Bentlin FRS, Pozebon D. Methodology for lanthanide elements quantification in NiMH batteries. Br J Anal Chem. 2012;8:358–365. [Google Scholar]
  • 7.Chandra D, Chien W-M, Talekar A. Metal hydrides for NiMH battery applications. Mater Matters. 2011;6(2):48–53. [Google Scholar]
  • 8.Ying TK, Gao XP, Hu WK, Wu F, Noréus D. Studies on rechargeable NiMH batteries. Int J Hydrogen Energy. 2006;31(4):525–530. [Google Scholar]
  • 9.Bäuerlein P, Antonius C, Löffler J, Kümpers J. Progress in high-power nickel–metal hydride batteries. J Power Sources. 2008;176(2):547–554. [Google Scholar]
  • 10.Fetcenko MA, et al. Recent advances in NiMH battery technology. J Power Sources. 2007;165(2):544–551. [Google Scholar]
  • 11.Binnemans K, Jones PT. Perspectives for the recovery of rare earths from end-of-life fluorescent lamps. J Rare Earths. 2014;32(3):195–200. [Google Scholar]
  • 12.Smets BMJ. Phosphors based on rare-earths, a new era in fluorescent lighting. Mater Chem Phys. 1987;16(3):283–299. [Google Scholar]
  • 13.Heidemann A, Hien S, Panofski E, Roll U. Compact fluorescent lamps. IEE Proc Sci Meas Tech. 1993;140(6):429–434. [Google Scholar]
  • 14.Hashimoto N, Takada Y, Sato K, Ibuki S. Green-luminescent (La,Ce)PO4:Tb phosphors for small size fluorescent lamps. J Lumin. 1991;48:893–897. [Google Scholar]
  • 15.Gupta CK, Krishnamurthy N. Extractive metallurgy of rare earths. Int Mater Rev. 1992;37(1):197–248. [Google Scholar]
  • 16.Alonso E, et al. Evaluating rare earth element availability: A case with revolutionary demand from clean technologies. Environ Sci Technol. 2012;46(6):3406–3414. doi: 10.1021/es203518d. [DOI] [PubMed] [Google Scholar]
  • 17.Humphries M. Rare Earth Elements: The Global Supply Chain. Congressional Research Service; Washington, DC: 2013. pp. 1–26. [Google Scholar]
  • 18.Hatch GP. Dynamics in the global market for rare earths. Elements. 2012;8(5):341–346. [Google Scholar]
  • 19.US Department of Energy 2011. Critical Materials Strategy (US Department of Energy, Washington, DC), pp 1–191.
  • 20.Sprecher B, et al. Life cycle inventory of the production of rare earths and the subsequent production of NdFeB rare earth permanent magnets. Environ Sci Technol. 2014;48(7):3951–3958. doi: 10.1021/es404596q. [DOI] [PubMed] [Google Scholar]
  • 21.Binnemans K, et al. Recycling of rare earths: A critical review. J Clean Prod. 2013;51(0):1–22. [Google Scholar]
  • 22.Reck BK, Graedel TE. Challenges in metal recycling. Science. 2012;337(6095):690–695. doi: 10.1126/science.1217501. [DOI] [PubMed] [Google Scholar]
  • 23.Darcy JW, et al. Challenges in recycling end-of-life rare earth magnets. JOM. 2013;65(11):1381–1382. [Google Scholar]
  • 24.Bandara HMD, Darcy JW, Apelian D, Emmert MH. Value analysis of neodymium content in shredder feed: Toward enabling the feasibility of rare earth magnet recycling. Environ Sci Technol. 2014;48(12):6553–6560. doi: 10.1021/es405104k. [DOI] [PubMed] [Google Scholar]
  • 25.Bandara HMD, Field KD, Emmert MH. Rare earth recovery from end-of-life motors employing green chemistry design principles. Green Chem. 2016;18(3):753–759. [Google Scholar]
  • 26.Bandara HMD, Mantell MA, Darcy JW, Emmert MH. Closing the lifecycle of rare earth magnets: Discovery of neodymium in slag from steel mills. Energy Technol. 2015;3(2):118–120. [Google Scholar]
  • 27.Bandara HMD, Mantell MA, Darcy JW, Emmert MH. Rare earth recycling: Forecast of recoverable Nd from shredder scrap and influence of recycling rates on price volatility. J Sustainable Metallurgy. 2015;1(3):179–188. [Google Scholar]
  • 28.Bogart JA, Lippincott CA, Carroll PJ, Schelter EJ. An operationally simple method for separating the rare-earth elements neodymium and dysprosium. Angew Chem Int Ed Engl. 2015;54(28):8222–8225. doi: 10.1002/anie.201501659. [DOI] [PubMed] [Google Scholar]
  • 29.Dupont D, Binnemans K. Rare-earth recycling using a functionalized ionic liquid for the selective dissolution and revalorization of Y2O3:Eu3+ from lamp phosphor waste. Green Chem. 2015;17(2):856–868. [Google Scholar]
  • 30.Schüler D, Buchert M, Liu R, Dittrich S, Merz C. Study on Rare Earths and Their Recycling. Öko-Institute e.V.; Darmstadt, Germany: 2011. [Google Scholar]
  • 31.Rabah MA. Recyclables recovery of europium and yttrium metals and some salts from spent fluorescent lamps. Waste Manag. 2008;28(2):318–325. doi: 10.1016/j.wasman.2007.02.006. [DOI] [PubMed] [Google Scholar]
  • 32.Mei G, Rao P, Mitsuaki M, Toyohisa F. 2009. Separation of red (Y2O3: Eu3+), blue (Sr, Ca, Ba)10(PO4)6Cl2: Eu2+ and green (LaPO4: Tb3+, Ce3+) rare earth phosphors by liquid/liquid extraction. J Wuhan Univ Technol, Mater Sci Ed 24(3):418–423.
  • 33.Takahashi T, et al. Separation and recovery of rare earth elements from phosphor sludge in processing plant of waste fluorescent lamp by pneumatic classification and sulfuric acidic leaching. Shigen to Sozai. 2001;117(7):579–585. [Google Scholar]
  • 34.Shimizu R, Sawada K, Enokida Y, Yamamoto I. Supercritical fluid extraction of rare earth elements from luminescent material in waste fluorescent lamps. J Supercrit Fluids. 2005;33(3):235–241. [Google Scholar]
  • 35.Takahashi T, et al. Synthesis of red phosphor (Y2O3:Eu3+) from waste phosphor sludge by coprecipitation process. Shigen to Sozai. 2002;118(5-6):413–418. [Google Scholar]
  • 36.Innocenzi V, De Michelis I, Kopacek B, Vegliò F. Yttrium recovery from primary and secondary sources: A review of main hydrometallurgical processes. Waste Manag. 2014;34(7):1237–1250. doi: 10.1016/j.wasman.2014.02.010. [DOI] [PubMed] [Google Scholar]
  • 37.Liu H, et al. Rare earth elements recycling from waste phosphor by dual hydrochloric acid dissolution. J Hazard Mater. 2014;272(0):96–101. doi: 10.1016/j.jhazmat.2014.02.043. [DOI] [PubMed] [Google Scholar]
  • 38.Tunsu C, Ekberg C, Retegan T. Characterization and leaching of real fluorescent lamp waste for the recovery of rare earth metals and mercury. Hydrometallurgy. 2014;144–145:91–98. [Google Scholar]
  • 39.Braconnier JJ, Rollat A. 2012. US patent publication US20120070351 A1 (March 22, 2012)
  • 40.Bogart JA, Lippincott CA, Carroll PJ, Booth CH, Schelter EJ. Controlled redox chemistry at cerium within a tripodal nitroxide ligand framework. Chemistry. 2015;21(49):17850–17859. doi: 10.1002/chem.201502952. [DOI] [PubMed] [Google Scholar]
  • 41.Kim JE, Carroll PJ, Schelter EJ. Bidentate nitroxide ligands stable toward oxidative redox cycling and their complexes with cerium and lanthanum. Chem Commun (Camb) 2015;51(81):15047–15050. doi: 10.1039/c5cc06052d. [DOI] [PubMed] [Google Scholar]
  • 42.Kim JE, Bogart JA, Carroll PJ, Schelter EJ, Schelter EJ. Rare earth metal complexes of bidentate nitroxide ligands: Synthesis and electrochemistry. Inorg Chem. 2016;55(2):775–784. doi: 10.1021/acs.inorgchem.5b02236. [DOI] [PubMed] [Google Scholar]
  • 43.Dorfner WL, Carroll PJ, Schelter EJ. A homoleptic [small eta]2 hydroxylaminato CeIV complex with S4 symmetry. Dalton Trans. 2014;43:6300–6303. doi: 10.1039/c4dt00287c. [DOI] [PubMed] [Google Scholar]
  • 44.Bogart JA, et al. Homoleptic cerium(III) and cerium(IV) nitroxide complexes: Significant stabilization of the 4+ oxidation state. Inorg Chem. 2013;52(19):11600–11607. doi: 10.1021/ic401974t. [DOI] [PubMed] [Google Scholar]
  • 45.Bogart JA, et al. A ligand field series for the 4f-block from experimental and DFT computed Ce(IV/III) electrochemical potentials. Inorg Chem. 2015;54(6):2830–2837. doi: 10.1021/ic503000z. [DOI] [PubMed] [Google Scholar]
  • 46.Gupta CK, Krishnamurthy N. Extractive Metallurgy of Rare Earths. CRC Press; New York: 2005. pp. 1–484. [Google Scholar]
  • 47.Yuan M, Luo A, Li D. Solvent extraction of lanthanides in aqueous nitrate media by Cyanex 302. Acta Metall Sin (Engl Lett) 1995;8:10–14. [Google Scholar]
  • 48.Haynes WM. CRC Handbook of Chemistry and Physics. 91st Ed. CRC Press; Boca Raton, FL: 2010. pp. 6-186–6-207. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary File
pnas.1612628113.sapp.pdf (2.3MB, pdf)

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES